Where Venture Capital Went 2024–2026: AI Takes Half the Money, Defense, Energy and the Barbell

Venture capital investment by sector from 2024 to 2026 is dominated by one category to a degree never seen before, and this article quantifies it year by year: how much was invested globally and by region, what share went to AI and to which layer of AI, which non-AI sectors (defence, energy, robotics, biotech, fintech) still attracted capital, how mega-rounds and mega-funds created a barbell, and what the 2026 year-to-date data say as of October 2026. It is the third chronological piece in the Sectors & Exits by Year pillar of the Kurums Venture Capital hub, following the 2015–2020 build-up and the 2021–2023 peak and reset.

How much venture capital was invested globally in 2024, 2025 and 2026 so far?

Global venture funding was about $314 billion in 2024 (Crunchbase), roughly $512 billion in 2025 (PitchBook) and $679 billion in the first three quarters of 2026 alone (Crunchbase, as of 2 October 2026), already the largest nine-month total on record.

The 2024 recovery was modest: Crunchbase counted close to $314 billion, up about 3% on 2023, with the US at $178 billion or 57% of the total and the San Francisco Bay Area alone at $90 billion. The year ended strongly, with Q4 2024 at $93 billion as Databricks ($10 billion at $62 billion), OpenAI ($6.6 billion at $157 billion), xAI, Waymo and Anthropic closed rounds; those five companies accounted for 43% of Q4 US deal value (PitchBook-NVCA). Billion-dollar rounds totalled $58.3 billion in 2024, 19% of all funding.

2025 was the second-largest year on record on PitchBook’s global series at about $512 billion, close to the 2021–2022 high-water mark, with AI above half of deal value and nearly a third of completed deals. Crunchbase’s AI tally for 2025 was $202.3 billion, up more than 75% on 2024’s $114 billion, with foundation-model companies alone raising $80 billion. SoftBank’s $40 billion investment in OpenAI was the largest deal of the year and Meta’s $14.3 billion investment in Scale AI the largest corporate strategic round; OpenAI and Anthropic together captured about 14% of global venture investment in 2025.

2026 has broken every quarterly record. Crunchbase reported roughly $308 billion in Q1 (first reported at $300 billion), $212 billion in Q2 and $159 billion in Q3, for $679 billion year to date; KPMG’s Venture Pulse, which uses a broader definition, puts the nine-month total at $728.3 billion with Q3 at $170.9 billion across 8,341 deals. In the US, PitchBook-NVCA counted $412.7 billion in H1 2026 alone, more than the whole of 2025, with Anthropic’s $65 billion round at a $965 billion post-money valuation the largest. How this capital eventually exits is tracked in our year-by-year table of the largest venture-backed exits.

Figures as published by each provider at the time of writing (October 2026); definitions differ, so Crunchbase and KPMG totals are not directly additive.

What share of venture capital went to AI, and to which layer?

AI took about a third of global venture funding in 2024, close to half in 2025 and between 64% and 80% of each quarter in 2026; within AI, foundation-model labs took about 40% of 2025 AI dollars, with infrastructure (compute, data centres, chips) and applications splitting the rest.

Crunchbase’s layer breakdown is the most useful. In 2025 foundation-model companies raised $80 billion, 40% of all AI funding, up from $31 billion and 27% in 2024; the rest went to AI infrastructure (CoreWeave before its IPO, Crusoe, Lambda, Nscale, Cerebras, Groq), data and tooling (Databricks, Scale AI) and applications in coding (Cursor, Cognition, Lovable), healthcare, legal, finance and defence. Some 58% of 2025 AI funding went into mega-rounds of $500 million or more, and corporate investors (SoftBank, Meta, Nvidia, Microsoft, Amazon, Google, ASML, Disney, Intercontinental Exchange) led a majority of the billion-dollar rounds. Among venture firms, Lightspeed, Founders Fund and Andreessen Horowitz led the most billion-dollar AI rounds in 2025.

In 2026 the concentration intensified. PitchBook-NVCA counted $355.9 billion of US AI deal value in H1 2026, 86% of all US venture dollars, with mega-deals of $100 million or more at 87.5% of the total. Q1 2026 was dominated by Anthropic’s $65 billion round and xAI’s $20 billion Series E at $230 billion (shortly before xAI merged into SpaceX at a $250 billion valuation). Q3 2026’s largest rounds were Databricks and Safe Superintelligence at $5 billion each, with Crusoe, Moonshot AI, Mistral, Nscale, Kling AI and The Boring Company each above $3 billion. The applications layer also matured: Cursor was reported at about $4 billion of annual recurring revenue by June 2026 before SpaceX agreed to acquire it for $60 billion, and Cognition raised $2 billion in Q3.

Where AI money sits geographically

The US took 79% of 2025 AI funding ($159 billion), and the Bay Area more than three-quarters of that ($122 billion). Europe’s AI champions are Mistral (France, $3.5 billion in Q3 2026 after its €1.7 billion 2025 round led by ASML), Helsing (Germany, $1.8 billion in Q3 2026), Nscale (UK) and Black Forest Labs; China’s are Moonshot AI ($3.5 billion), Kling AI ($3 billion), Zhipu and DeepSeek’s ecosystem, which pushed China’s Q3 2026 venture total to $27.2 billion of Asia’s $39.8 billion (KPMG).

Which non-AI sectors still attracted capital: defence, energy, robotics, biotech?

Defence and aerospace, energy and data-centre power, robotics and physical AI, biotech and a recovering fintech were the non-AI winners; KPMG reports that aerospace, robotics, data centres, semiconductors and energy each raised $10 billion or more in Q3 2026 alone.

Defence tech went from fringe to core. Anduril raised $2.5 billion at $30.5 billion in June 2025 and continued to scale; Helsing raised €600 million at €12 billion in 2025 and $1.8 billion in Q3 2026; Quantum Systems raised $1.2 billion in Q3 2026; Saronic, Shield AI, Castelion and Chaos Industries raised nine-figure rounds; and European governments’ rearmament pledges after 2022 turned defence into one of Europe’s few mega-round categories. Space and launch (SpaceX’s pre-IPO rounds, Stoke, Relativity, Isar) sit in the same bucket.

Energy became an AI derivative. Data-centre power demand pushed capital into fusion (Commonwealth Fusion Systems’ $863 million round in August 2025, Helion, TAE), small modular and advanced nuclear (X-energy, Oklo’s public listing, Radiant), geothermal (Fervo), grid software and batteries, while Crusoe ($3.9 billion in Q3 2026) and Nscale raised to build AI-dedicated power and compute campuses. Firmus Technologies in Australia raised $2 billion for the same purpose. Robotics and “physical AI” followed: Figure raised more than $1 billion at $39 billion in September 2025, Physical Intelligence, Skild and Apptronik raised large rounds, and Unitree’s IPO at a $9 billion valuation in Q3 2026 showed the public market was open to the category.

Biotech recovered on the back of AI-driven discovery (Isomorphic, Xaira, Lila) and a reopened IPO window, with 13 biotech IPOs raising $4.5 billion in H1 2026, more than the whole of 2025 (Stout). Fintech came back as an exit story first, with Chime, Circle and Klarna listing in 2025, then as a funding story, led by neobanks in Latin America and by stablecoin and payments infrastructure after Stripe’s acquisitions. Climate tech, the 2021–2023 resilience story, slowed in 2024–2025 as US policy shifted and capital rotated to energy for AI; the categories that kept raising were those adjacent to data-centre demand. Founders in these sectors should read how founders outside AI can still raise in 2026 for the practical positioning that works.

Round figures as reported by Crunchbase, KPMG Venture Pulse, PitchBook-NVCA and company announcements, 2024 to October 2026.

How did mega-rounds and mega-funds create the barbell?

Both ends of the market grew while the middle shrank: 27 companies raised $1 billion or more in Q3 2026 (a record, taking about a third of the quarter’s capital), seed funding held at about $13 billion a quarter, but deals under $100 million fell to 12.5% of US deal value in H1 2026 from 43.8% in 2024.

The deal-size data are stark. PitchBook-NVCA reports that US deals under $100 million totalled just $51.4 billion in H1 2026, a 12.5% share, down from 33.1% in 2025 and 43.8% in 2024, while seven rounds of $1 billion or more in Q2 2026 alone totalled $87.2 billion. Crunchbase’s Q3 2026 stage split shows late-stage at $105 billion, early-stage at $40.6 billion (half of it in rounds of $100 million-plus) and seed at $13 billion. The pattern is exactly the one described in our analysis of the venture capital barbell effect: a company is either a presumed category winner raising at any price or it is raising a modest seed round, and the Series B-to-D middle outside AI is where fundraising is hardest.

Fundraising mirrors deployment. Global VC fundraising fell to $118.6 billion in 2025, a decade low and nearly $100 billion less than 2024 (PitchBook), with 537 funds closed on $66.1 billion compared with 1,777 funds on $222.9 billion in 2022. Yet H1 2026 US fundraising reached $72.4 billion across 405 funds, almost matching full-year 2025, because three firms raised $34.8 billion between them: Andreessen Horowitz $14.2 billion across seven funds, Founders Fund $10.6 billion and Thrive Capital $10 billion, or 48.1% of H1 commitments. Experienced firms took 89% of capital; first-time funds raised just $3.4 billion across 53 vehicles. The median time to close a fund fell to 6.4 months for those that could raise at all. Our review of the largest US venture firms and their 2024–2026 funds covers who those managers are.

How did the regional split change: US dominance, Europe and Asia?

The US took 57% of global venture funding in 2024, about 57% again in Q3 2026 after peaking near 83% in Q1 2026 (Crunchbase), while Europe grew for eight straight quarters to $28.4 billion in Q3 2026 and Asia was $39.8 billion, of which China was $27.2 billion (KPMG).

US dominance is an AI artefact. In quarters when OpenAI, Anthropic or xAI close a round, the US share spikes (Q1 2026 at about 83%); in quarters when European and Chinese labs raise, it falls back toward the high 50s. The San Francisco Bay Area alone took 24% of global venture funding in Q3 2026 and $122 billion of 2025 AI funding. Outside AI, US venture is not obviously larger than it was in 2019–2020, which is why the PitchBook-NVCA data on first-time financings (an estimated 5,674 in H1 2026, on pace for a record year) are more encouraging than the headline share suggests.

Europe is the quiet success. KPMG records eight consecutive quarters of growth to a 16-quarter high of $28.4 billion in Q3 2026, led by Mistral, Helsing, Quantum Systems, Nscale and Bending Spoons (which listed at an $18.4 billion valuation), with the UK at $6.8 billion; PitchBook’s 2025 European total was about €66 billion, flat on 2024. Europe’s weakness remains exits, as our analysis of Germany’s scaleup gap documents. Asia is split: China’s AI and semiconductor rounds (Moonshot, Kling, Enflame, CXMT’s $8.6 billion IPO) keep totals high, Japan is small at $1.2 billion in Q3 2026 (see why Japan has so few unicorns), and India fell to a six-quarter low of $2.6 billion. Canada’s $866 million was its lowest quarter since 2018, and Latin America’s activity is concentrated in Brazilian and Mexican fintech.

What do the 2026 year-to-date numbers say as of October 2026?

As of October 2026, 2026 is already the largest venture year ever on every dataset: $679 billion through Q3 on Crunchbase, $728.3 billion on KPMG, with Q3 down 25% from Q2 but up 53% year on year, and corporate venture investment alone at $532.8 billion year to date, above any previous full year.

The quarterly trajectory matters. Q1 2026’s roughly $308 billion was inflated by Anthropic and xAI; Q2’s $212 billion included seven US rounds of $1 billion-plus; Q3’s $159 billion was lower in total but set a record for the number of billion-dollar rounds (27) and showed the broadest sector spread of the cycle, with aerospace, robotics, data centres, semiconductors and energy each above $10 billion. Non-AI funding in Q3 was roughly $57 billion, which on its own would have been a healthy quarter by 2023 standards. Exit markets reopened in parallel: SpaceX’s $75 billion IPO in June 2026, the Cerebras IPO, the SpaceX–xAI merger at $250 billion and Nvidia’s $12.9 billion bid for Hugging Face are examined in our analysis of IPO, M&A and secondary exit routes.

Two signals deserve caution. First, Crunchbase notes that AI’s share fell from about 80% in Q1 to 64% in Q3 while absolute totals also fell, which could mean either broadening or a plateau in lab financing now that OpenAI and Anthropic have filed confidentially for IPOs. Second, fundraising is concentrated enough that a single LP retreat from the top three firms would show up immediately in deal value. For LPs, the relevant reading is which venture funds are in market in 2026.

What does the 2024–2026 allocation mean for founders and LPs?

For founders it means the sector premium is real and widening: AI companies raise faster, larger and at higher prices, while non-AI companies must show public-market-grade metrics. For LPs it means concentration risk in both deployment and fund selection is at a historical high.

Founders outside AI have three realistic strategies: attach to the AI demand chain (energy, chips, data, robotics, security), present as an AI-native business with proof of revenue rather than a wrapper, or raise from sector-specialist and regional funds that do not benchmark against Bay Area AI rounds. The pitch materials that work in this market are covered in what a strong investor pitch actually contains. AI founders face the opposite problem: pricing. A $1 billion round at a $30 billion valuation requires an exit above $60–100 billion to return a fund, and 2026’s exit data show how few companies reach that bar.

LPs should note that 2025 global fundraising ($118.6 billion) was the lowest in a decade while deployment approached a record, which means the industry is drawing down 2021–2022 dry powder and corporate capital rather than fresh LP commitments. Distributions remain thin for 2019–2021 vintages, which is why the DPI data in our exits-by-year analysis are more decision-relevant than headline AUM. Browse the full Sectors & Exits by Year pillar for the earlier instalments and the exit-route analysis.

Frequently Asked Questions